Species reconfigured

src/mmaseq/config/species_configs/CPO.yaml

@@ -0,0 +1,8 @@
+plasmidfinder:
+  options: "-l 80 -t 80"
+  assembler: shovill
+
+blastn:
+  options: "-perc_identity 99.0"
+  assembler: shovill
+  database : OXAndm

src/mmaseq/config/species_configs/C_coli.yaml

@@ -0,0 +1,12 @@
+plasmidfinder:
+  options: "-l 80 -t 80"
+  assembler: shovill
+
+blastn:
+  options: "-perc_identity 99.0"
+  assembler: shovill
+  database : OXAndm
+
+pointfinder:
+    options: --species 'Campylobacter coli'
+    assembler: shovill

src/mmaseq/config/species_configs/C_jejuni.yaml

@@ -0,0 +1,12 @@
+plasmidfinder:
+  options: "-l 80 -t 80"
+  assembler: shovill
+
+blastn:
+  options: "-perc_identity 99.0"
+  assembler: shovill
+  database : OXAndm
+
+pointfinder:
+    options: --species 'Campylobacter jejuni'
+    assembler: shovill

src/mmaseq/config/species_configs/Campylobacter.yaml

@@ -0,0 +1,12 @@
+plasmidfinder:
+  options: "-l 80 -t 80"
+  assembler: shovill
+
+blastn:
+  options: "-perc_identity 99.0"
+  assembler: shovill
+  database : OXAndm
+
+pointfinder:
+    options: --species 'Campylobacter coli'
+    assembler: shovill

src/mmaseq/config/species_configs/E_Faecalis.yaml

@@ -1,18 +1,15 @@
 plasmidfinder:
-    reads: True
-
-virulencefinder:
-    reads: True
-
-serotypefinder:
-    reads: True
+  options: "-l 80 -t 80"
+  assembler: shovill
 
-amrfinder:
-    options: --organism 'Enterococcus_faecalis'
-    assembler: [shovill]  
+blastn:
+  options: "-perc_identity 99.0"
+  assembler: shovill
+  database : OXAndm
 
-mlst:
-    assembler: [shovill]  
+pointfinder:
+    options: --species 'Enterococcus faecalis'
+    assembler: shovill
 
 lrefinder:
     database : [elmDB]

src/mmaseq/config/species_configs/E_Faecium.yaml

@@ -1,23 +1,15 @@
 plasmidfinder:
-    reads: True
-
-virulencefinder:
-    reads: True
+  options: "-l 80 -t 80"
+  assembler: shovill
 
-serotypefinder:
-    reads: True
+blastn:
+  options: "-perc_identity 99.0"
+  assembler: shovill
+  database : OXAndm
 
-resfinder:
-    reads: False
+pointfinder:
+    options: --species 'Enterococcus faecium'
     assembler: shovill
-    options: --species 'Other'
-
-amrfinder:
-    options: --organism 'Enterococcus_faecium'
-    assembler: [shovill]  
-
-mlst:
-    assembler: [shovill]  
 
 lrefinder:
     database : [elmDB]

src/mmaseq/config/species_configs/E_coli.yaml

@@ -1,45 +1,36 @@
-kmeraligner_wrangler:
-    options: --organism 'Escherichia coli'
-    database: ecoligenes
-    reads: True
-
 resfinder:
     options: --species 'Escherichia coli'
-    reads: True
+    assembler: shovill
 
 pointfinder:
     options: --species 'Escherichia coli'
-    reads: True
+    assembler: shovill
 
-disinfinder:
-    options: --species 'Escherichia coli'
-    reads: True
+plasmidfinder:
+  options: "-l 80 -t 80"
+  assembler: shovill
 
-plasmidfinder: 
-    reads: True
+blastn:
+  options: "-perc_identity 99.0"
+  assembler: shovill
+  database : OXAndm
 
 virulencefinder: 
-    reads: True
+    assembler: shovill
 
-serotypefinder: 
+serotypefinder:
     reads: True
 
-amrfinder:
-    options: --organism 'Escherichia'
-    assembler: shovill 
+kmeraligner_wrangler:
+    options: --organism 'Escherichia coli'
+    database: ecoligenes
+    reads: True
 
 chtyper:
     database: fumCH 
     reads: True
 
-mlst:
-    assembler: shovill  
-
 blastn:
     options: "-perc_identity 90.0"
     assembler: shovill
     database : OXAndm
-
-kleborate:
-    options: --preset escherichia
-    assembler: shovill

src/mmaseq/config/species_configs/H_pylori.yaml

@@ -0,0 +1,12 @@
+plasmidfinder:
+  options: "-l 80 -t 80"
+  assembler: shovill
+
+blastn:
+  options: "-perc_identity 99.0"
+  assembler: shovill
+  database : OXAndm
+
+pointfinder:
+    options: --species 'Helicobacter pylori'
+    assembler: shovill

src/mmaseq/config/species_configs/K_oxytoca.yaml

@@ -0,0 +1,37 @@
+plasmidfinder:
+  options: "-l 80 -t 80"
+  assembler: shovill
+
+blastn:
+  options: "-perc_identity 99.0"
+  assembler: shovill
+  database : OXAndm
+
+resfinder:
+    options: --species 'Klebsiella'
+    reads: True
+
+pointfinder:
+    options: --species 'Klebsiella'
+    reads: True
+
+disinfinder:
+    options: --species 'Klebsiella'
+    reads: True
+
+virulencefinder:
+    reads: True
+
+serotypefinder:
+    reads: True
+
+amrfinder:
+    options: --organism 'Klebsiella_oxytoca'
+    assembler: shovill 
+
+kleborate:
+    options: --preset kosc
+    assembler: shovill
+
+mlst:
+    assembler: shovill  

src/mmaseq/config/species_configs/K_pneumoniae.yaml

@@ -1,5 +1,11 @@
 plasmidfinder:
-    reads: True
+  options: "-l 80 -t 80"
+  assembler: shovill
+
+blastn:
+  options: "-perc_identity 99.0"
+  assembler: shovill
+  database : OXAndm
 
 resfinder:
     options: --species 'Klebsiella'

src/mmaseq/config/species_configs/M_tuberculosis.yaml

@@ -0,0 +1,8 @@
+plasmidfinder:
+  options: "-l 80 -t 80"
+  assembler: shovill
+
+blastn:
+  options: "-perc_identity 99.0"
+  assembler: shovill
+  database : OXAndm

src/mmaseq/config/species_configs/N_gonorrhoeae.yaml

@@ -0,0 +1,12 @@
+plasmidfinder:
+  options: "-l 80 -t 80"
+  assembler: shovill
+
+blastn:
+  options: "-perc_identity 99.0"
+  assembler: shovill
+  database : OXAndm
+
+pointfinder:
+    options: --species 'Neisseria gonorrhoeae'
+    assembler: shovill

src/mmaseq/config/species_configs/N_meningitidis.yaml

@@ -1,5 +1,11 @@
 plasmidfinder:
-    reads: True
+  options: "-l 80 -t 80"
+  assembler: shovill
+
+blastn:
+  options: "-perc_identity 99.0"
+  assembler: shovill
+  database : OXAndm
 
 virulencefinder:
     reads: True

src/mmaseq/config/species_configs/S_aureus.yaml

@@ -1,2 +1,2 @@
 spatyper:
-  assembler: [spades]
+  assembler: [spades]

src/mmaseq/config/species_configs/Salmonella.yaml

@@ -0,0 +1,23 @@
+pointfinder:
+    options: --species 'Salmonella'
+    assembler: shovill
+
+plasmidfinder:
+  options: "-l 80 -t 80"
+  assembler: shovill
+
+blastn:
+  options: "-perc_identity 99.0"
+  assembler: shovill
+  database : OXAndm
+
+amrfinder:
+  options: --organism 'Salmonella'
+  assembler: shovill
+
+seqsero2: 
+  reads: True
+
+sistr:
+  assembler: shovill
+

src/mmaseq/config/species_configs/default.yaml

@@ -7,20 +7,24 @@ resfinder:
 
 plasmidfinder:
     options: ""
+    reads: True
 
 virulencefinder:
     options: ""
+    reads: True
 
 serotypefinder:
     options: ""
-
+    reads: True
+
 amrfinder:
     assembler: shovill
     options: ""
-    
+
 meningotype:
     assembler: shovill
 
 kmeraligner:
     database: elmDB
-    options : -ID 80 -1t1 -cge
+    options : -ID 80 -1t1 -cge
+    reads: True

src/mmaseq/data/samplesheet.tsv

@@ -2,7 +2,7 @@ sample_name	read1	read2	assembly	config
 ERR3528110	reads/ERR3528110_1.fastq.gz	reads/ERR3528110_2.fastq.gz	assemblies/ERR3528110.fasta	E_coli.yaml
 SRR4046826	reads/SRR4046826_1.fastq.gz	reads/SRR4046826_2.fastq.gz	assemblies/SRR4046826.fasta	K_pneumoniae.yaml
 SRR25448586	reads/SRR25448586_1.fastq.gz	reads/SRR25448586_2.fastq.gz	assemblies/SRR25448586.fasta	N_meningitidis.yaml
-SRR26205262	reads/SRR26205262_1.fastq.gz	reads/SRR26205262_2.fastq.gz	assemblies/SRR26205262.fasta	S_enterica.yaml
+SRR26205262	reads/SRR26205262_1.fastq.gz	reads/SRR26205262_2.fastq.gz	assemblies/SRR26205262.fasta	Salmonella.yaml
 ERR2929615	reads/ERR2929615_1.fastq.gz	reads/ERR2929615_2.fastq.gz	assemblies/ERR2929615.fasta	E_Faecalis.yaml
 ERR2929578	reads/ERR2929578_1.fastq.gz	reads/ERR2929578_2.fastq.gz	assemblies/ERR2929578.fasta	E_Faecium.yaml
 ERR142064	reads/ERR142064_1.fastq.gz	reads/ERR142064_2.fastq.gz	assemblies/ERR142064.fasta	C_difficile.yaml

src/mmaseq/utils/sample_config.py

@@ -87,21 +87,23 @@ def determine_sample_configs(samplesheet, config_dir, ignore_assemblies):
 
         # Handle missing configuration file
         if not os.path.isfile(cfg_path):
-            print(
-                f"Warning: Config file specified in samplesheet {cfg} "
-                f"does not exist in {config_dir}!"
+            available = sorted(p.name for p in Path(config_dir).glob("*.yaml"))
+            logger.warning(
+                f"Sample '{sample}' references config '{cfg}' which was not "
+                f"found in {config_dir}. Available configs: "
+                f"{', '.join(available) if available else '(none)'}."
             )
             cfg_path = None
 
             default_path = f"{config_dir}/default.yaml"
 
             # Ensure that default file exists and use it
             if os.path.exists(default_path):
-                print("Using default.yaml instead")
+                logger.warning(f"Falling back to default.yaml for sample '{sample}'.")
                 cfg_path = default_path
             else:
-                print(
-                    "Warning: Default configuration file is missing, "
+                logger.error(
+                    "Default configuration file is missing, "
                     "please recreate it to enable default analysis: "
                     f"{default_path}"
                 )